Re: [EXT] Re: [EXT] Re: [EXT] Re: [EXT] Re: [EXT] Re: Beam Dataframe - sort and grouping

[Wenbing Bai]

Thank you for the clarification! Is there a way to control the number of
shards, i.e. the bundle? I know in pure Beam IO connectors, we have
num_shards supported, for example, WriteToParquet

.

Wenbing

On Thu, May 13, 2021 at 4:35 PM Robert Bradshaw  wrote:

> Sharding is determined by the distribution of work. Each worker writes to
> its own shard, and in the case of dynamic partitioning, etc. workers may
> end up processing more than one "bundle" of items and hence produce more
> than one shard. See also
> https://beam.apache.org/documentation/runtime/model/
>
> On Thu, May 13, 2021 at 3:58 PM Wenbing Bai 
> wrote:
>
>> Hi team,
>>
>> I have another question when using Beam Dataframe IO connector. I tried
>> to_parquet, and my data are written to several different files. I am
>> wondering how I can control the number of files (shards) or how the
>> sharding is done for to_parquet and other Beam Dataframe IO APIs?
>>
>> Thank you!
>> Wenbing
>>
>> On Tue, May 11, 2021 at 12:20 PM Kenneth Knowles  wrote:
>>
>>> +dev 
>>>
>>> In the Beam Java ecosystem, this functionality is provided by the Sorter
>>> library (
>>> https://beam.apache.org/documentation/sdks/java-extensions/#sorter).
>>> I'm curious what people think about various options:
>>>
>>>  - Python version of the transform(s)
>>>  - Expose sorter as xlang transform(s)
>>>  - Convenience transforms (that use pandas in DoFns?) to just do it for
>>> small data per key to achieve compatibility
>>>  - Beam model extension so that runners can do it as part of GBK
>>>
>>> Kenn
>>>
>>> On Mon, May 10, 2021 at 5:26 PM Wenbing Bai 
>>> wrote:
>>>
 Hi Robert and Brian,

 I don't know why I didn't catch your replies. But thank you so much for
 looking at this.

 My parquet files will be consumed by downstreaming processes which
 require data points with the same "key1" that are sorted by "key2". The
 downstreaming process, for example, will make a rolling window with size N
 that reads N records together at one time. But note, the rolling window
 will not cross different "key1".

 So that is saying, 1) I don't need to sort the whole dataset. 2) all
 data with the same "key1" should be located together.

 I am not sure if I explain the use case clearly. Let me know what you
 think.

 Wenbing


 On Tue, Apr 20, 2021 at 5:01 PM Robert Bradshaw 
 wrote:

> It would also be helpful to understand what your overall objective is
> with this output. Is there a reason you need it sorted/partitioned in a
> certain way?
>
> On Tue, Apr 20, 2021 at 4:51 PM Brian Hulette 
> wrote:
>
>> Hi Wenbing,
>> Sorry for taking so long to get back to you on this.
>> I discussed this with Robert offline and we came up with a potential
>> workaround - you could try writing out the Parquet file from within the
>> groupby.apply method. You can use beam's FileSystems abstraction to open 
>> a
>> Python file object referencing a cloud storage file, and pass that file
>> object directly to the pandas to_parquet. It would look something like 
>> this:
>>
>>   df.groupby('key1').apply(lambda df:
>> df.sort_values(by='key2').to_parquet(FileSystems.open("gs://bucket/file.pq"))
>>
>> If writing out sorted, partitioned parquet files is a common use-case
>> we should think about making this easier though. At the very least
>> partition_cols should work, I filed BEAM-12201 [1] for this. That alone
>> won't be enough as our implementation will likely reshuffle the dataset 
>> to
>> enforce the partitioning, removing any sorting that you've applied, so 
>> we'd
>> also need to think about how to optimize the pipeline to avoid that 
>> shuffle.
>>
>> Brian
>>
>> [1] https://issues.apache.org/jira/browse/BEAM-12201
>>
>> On Wed, Apr 7, 2021 at 9:02 PM Wenbing Bai 
>> wrote:
>>
>>> Thank you, Brian. I tried `partition_cols`, but it is not working. I
>>> tried pure pandas, it does work, so I am not sure if anything wrong with
>>> Beam.
>>>
>>> Wenbing
>>>
>>> On Wed, Apr 7, 2021 at 2:56 PM Brian Hulette 
>>> wrote:
>>>
 Hm, to_parquet does have a `partition_cols` argument [1] which we
 pass through [2]. It would be interesting to see what
 `partition_cols='key1'` does - I suspect it won't work perfectly 
 though.

 Do you have any thoughts here Robert?

 [1]
 https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_parquet.html
 [2]
 https://github.com/apache/beam/blob/a8cd05932bed9b2480316fb8518409636cb2733b/sdks/python/apache_beam/dataframe/io.py#L525

 On Wed, Apr 7, 2021 at 2:22 PM 

Re: [EXT] Re: [EXT] Re: [EXT] Re: [EXT] Re: Beam Dataframe - sort and grouping

[Robert Bradshaw]

Sharding is determined by the distribution of work. Each worker writes to
its own shard, and in the case of dynamic partitioning, etc. workers may
end up processing more than one "bundle" of items and hence produce more
than one shard. See also
https://beam.apache.org/documentation/runtime/model/

On Thu, May 13, 2021 at 3:58 PM Wenbing Bai 
wrote:

> Hi team,
>
> I have another question when using Beam Dataframe IO connector. I tried
> to_parquet, and my data are written to several different files. I am
> wondering how I can control the number of files (shards) or how the
> sharding is done for to_parquet and other Beam Dataframe IO APIs?
>
> Thank you!
> Wenbing
>
> On Tue, May 11, 2021 at 12:20 PM Kenneth Knowles  wrote:
>
>> +dev 
>>
>> In the Beam Java ecosystem, this functionality is provided by the Sorter
>> library (
>> https://beam.apache.org/documentation/sdks/java-extensions/#sorter). I'm
>> curious what people think about various options:
>>
>>  - Python version of the transform(s)
>>  - Expose sorter as xlang transform(s)
>>  - Convenience transforms (that use pandas in DoFns?) to just do it for
>> small data per key to achieve compatibility
>>  - Beam model extension so that runners can do it as part of GBK
>>
>> Kenn
>>
>> On Mon, May 10, 2021 at 5:26 PM Wenbing Bai 
>> wrote:
>>
>>> Hi Robert and Brian,
>>>
>>> I don't know why I didn't catch your replies. But thank you so much for
>>> looking at this.
>>>
>>> My parquet files will be consumed by downstreaming processes which
>>> require data points with the same "key1" that are sorted by "key2". The
>>> downstreaming process, for example, will make a rolling window with size N
>>> that reads N records together at one time. But note, the rolling window
>>> will not cross different "key1".
>>>
>>> So that is saying, 1) I don't need to sort the whole dataset. 2) all
>>> data with the same "key1" should be located together.
>>>
>>> I am not sure if I explain the use case clearly. Let me know what you
>>> think.
>>>
>>> Wenbing
>>>
>>>
>>> On Tue, Apr 20, 2021 at 5:01 PM Robert Bradshaw 
>>> wrote:
>>>
 It would also be helpful to understand what your overall objective is
 with this output. Is there a reason you need it sorted/partitioned in a
 certain way?

 On Tue, Apr 20, 2021 at 4:51 PM Brian Hulette 
 wrote:

> Hi Wenbing,
> Sorry for taking so long to get back to you on this.
> I discussed this with Robert offline and we came up with a potential
> workaround - you could try writing out the Parquet file from within the
> groupby.apply method. You can use beam's FileSystems abstraction to open a
> Python file object referencing a cloud storage file, and pass that file
> object directly to the pandas to_parquet. It would look something like 
> this:
>
>   df.groupby('key1').apply(lambda df:
> df.sort_values(by='key2').to_parquet(FileSystems.open("gs://bucket/file.pq"))
>
> If writing out sorted, partitioned parquet files is a common use-case
> we should think about making this easier though. At the very least
> partition_cols should work, I filed BEAM-12201 [1] for this. That alone
> won't be enough as our implementation will likely reshuffle the dataset to
> enforce the partitioning, removing any sorting that you've applied, so 
> we'd
> also need to think about how to optimize the pipeline to avoid that 
> shuffle.
>
> Brian
>
> [1] https://issues.apache.org/jira/browse/BEAM-12201
>
> On Wed, Apr 7, 2021 at 9:02 PM Wenbing Bai 
> wrote:
>
>> Thank you, Brian. I tried `partition_cols`, but it is not working. I
>> tried pure pandas, it does work, so I am not sure if anything wrong with
>> Beam.
>>
>> Wenbing
>>
>> On Wed, Apr 7, 2021 at 2:56 PM Brian Hulette 
>> wrote:
>>
>>> Hm, to_parquet does have a `partition_cols` argument [1] which we
>>> pass through [2]. It would be interesting to see what
>>> `partition_cols='key1'` does - I suspect it won't work perfectly though.
>>>
>>> Do you have any thoughts here Robert?
>>>
>>> [1]
>>> https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_parquet.html
>>> [2]
>>> https://github.com/apache/beam/blob/a8cd05932bed9b2480316fb8518409636cb2733b/sdks/python/apache_beam/dataframe/io.py#L525
>>>
>>> On Wed, Apr 7, 2021 at 2:22 PM Wenbing Bai <
>>> wenbing@getcruise.com> wrote:
>>>
 Hi Robert and Brian,

 I tried groupby in my case. Here is my pipeline code. I do see all
 the data in the final parquet file are sorted in each group. However, 
 I'd
 like to write each partition (group) to an individual file, how can I
 achieve it? In addition, I am using the master of Apache Beam SDK, how 
 can
 I test the pipeline with DataflowRunner considering there is no 
 dataflow

Re: [EXT] Re: [EXT] Re: [EXT] Re: [EXT] Re: Beam Dataframe - sort and grouping

[Wenbing Bai]

Hi team,

I have another question when using Beam Dataframe IO connector. I tried
to_parquet, and my data are written to several different files. I am
wondering how I can control the number of files (shards) or how the
sharding is done for to_parquet and other Beam Dataframe IO APIs?

Thank you!
Wenbing

On Tue, May 11, 2021 at 12:20 PM Kenneth Knowles  wrote:

> +dev 
>
> In the Beam Java ecosystem, this functionality is provided by the Sorter
> library (
> https://beam.apache.org/documentation/sdks/java-extensions/#sorter). I'm
> curious what people think about various options:
>
>  - Python version of the transform(s)
>  - Expose sorter as xlang transform(s)
>  - Convenience transforms (that use pandas in DoFns?) to just do it for
> small data per key to achieve compatibility
>  - Beam model extension so that runners can do it as part of GBK
>
> Kenn
>
> On Mon, May 10, 2021 at 5:26 PM Wenbing Bai 
> wrote:
>
>> Hi Robert and Brian,
>>
>> I don't know why I didn't catch your replies. But thank you so much for
>> looking at this.
>>
>> My parquet files will be consumed by downstreaming processes which
>> require data points with the same "key1" that are sorted by "key2". The
>> downstreaming process, for example, will make a rolling window with size N
>> that reads N records together at one time. But note, the rolling window
>> will not cross different "key1".
>>
>> So that is saying, 1) I don't need to sort the whole dataset. 2) all data
>> with the same "key1" should be located together.
>>
>> I am not sure if I explain the use case clearly. Let me know what you
>> think.
>>
>> Wenbing
>>
>>
>> On Tue, Apr 20, 2021 at 5:01 PM Robert Bradshaw 
>> wrote:
>>
>>> It would also be helpful to understand what your overall objective is
>>> with this output. Is there a reason you need it sorted/partitioned in a
>>> certain way?
>>>
>>> On Tue, Apr 20, 2021 at 4:51 PM Brian Hulette 
>>> wrote:
>>>
 Hi Wenbing,
 Sorry for taking so long to get back to you on this.
 I discussed this with Robert offline and we came up with a potential
 workaround - you could try writing out the Parquet file from within the
 groupby.apply method. You can use beam's FileSystems abstraction to open a
 Python file object referencing a cloud storage file, and pass that file
 object directly to the pandas to_parquet. It would look something like 
 this:

   df.groupby('key1').apply(lambda df:
 df.sort_values(by='key2').to_parquet(FileSystems.open("gs://bucket/file.pq"))

 If writing out sorted, partitioned parquet files is a common use-case
 we should think about making this easier though. At the very least
 partition_cols should work, I filed BEAM-12201 [1] for this. That alone
 won't be enough as our implementation will likely reshuffle the dataset to
 enforce the partitioning, removing any sorting that you've applied, so we'd
 also need to think about how to optimize the pipeline to avoid that 
 shuffle.

 Brian

 [1] https://issues.apache.org/jira/browse/BEAM-12201

 On Wed, Apr 7, 2021 at 9:02 PM Wenbing Bai 
 wrote:

> Thank you, Brian. I tried `partition_cols`, but it is not working. I
> tried pure pandas, it does work, so I am not sure if anything wrong with
> Beam.
>
> Wenbing
>
> On Wed, Apr 7, 2021 at 2:56 PM Brian Hulette 
> wrote:
>
>> Hm, to_parquet does have a `partition_cols` argument [1] which we
>> pass through [2]. It would be interesting to see what
>> `partition_cols='key1'` does - I suspect it won't work perfectly though.
>>
>> Do you have any thoughts here Robert?
>>
>> [1]
>> https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_parquet.html
>> [2]
>> https://github.com/apache/beam/blob/a8cd05932bed9b2480316fb8518409636cb2733b/sdks/python/apache_beam/dataframe/io.py#L525
>>
>> On Wed, Apr 7, 2021 at 2:22 PM Wenbing Bai 
>> wrote:
>>
>>> Hi Robert and Brian,
>>>
>>> I tried groupby in my case. Here is my pipeline code. I do see all
>>> the data in the final parquet file are sorted in each group. However, 
>>> I'd
>>> like to write each partition (group) to an individual file, how can I
>>> achieve it? In addition, I am using the master of Apache Beam SDK, how 
>>> can
>>> I test the pipeline with DataflowRunner considering there is no dataflow
>>> worker image available?
>>>
>>> data = [
>>> {
>>> "key1": 1000 + i % 10,
>>> "key2": randrange(1),
>>> "feature_1": "somestring{}".format(i)
>>> } for i in range(1)
>>> ]
>>>
>>> class TestRow(typing.NamedTuple):
>>> key1: int
>>> key2: int
>>> feature_1: str
>>>
>>> with beam.Pipeline() as p:
>>> pcoll = (
>>> p
>>> | beam.Create(data)
>>> | beam.Map(lambda x:x).with_output_types(TestRow)
>>> )
>>>
>>> df = 

Re: [EXT] Re: [EXT] Re: [EXT] Re: Beam Dataframe - sort and grouping

[Wenbing Bai]

Hi Robert and Brian,

I don't know why I didn't catch your replies. But thank you so much for
looking at this.

My parquet files will be consumed by downstreaming processes which require
data points with the same "key1" that are sorted by "key2". The
downstreaming process, for example, will make a rolling window with size N
that reads N records together at one time. But note, the rolling window
will not cross different "key1".

So that is saying, 1) I don't need to sort the whole dataset. 2) all data
with the same "key1" should be located together.

I am not sure if I explain the use case clearly. Let me know what you think.

Wenbing


On Tue, Apr 20, 2021 at 5:01 PM Robert Bradshaw  wrote:

> It would also be helpful to understand what your overall objective is with
> this output. Is there a reason you need it sorted/partitioned in a
> certain way?
>
> On Tue, Apr 20, 2021 at 4:51 PM Brian Hulette  wrote:
>
>> Hi Wenbing,
>> Sorry for taking so long to get back to you on this.
>> I discussed this with Robert offline and we came up with a potential
>> workaround - you could try writing out the Parquet file from within the
>> groupby.apply method. You can use beam's FileSystems abstraction to open a
>> Python file object referencing a cloud storage file, and pass that file
>> object directly to the pandas to_parquet. It would look something like this:
>>
>>   df.groupby('key1').apply(lambda df:
>> df.sort_values(by='key2').to_parquet(FileSystems.open("gs://bucket/file.pq"))
>>
>> If writing out sorted, partitioned parquet files is a common use-case we
>> should think about making this easier though. At the very least
>> partition_cols should work, I filed BEAM-12201 [1] for this. That alone
>> won't be enough as our implementation will likely reshuffle the dataset to
>> enforce the partitioning, removing any sorting that you've applied, so we'd
>> also need to think about how to optimize the pipeline to avoid that shuffle.
>>
>> Brian
>>
>> [1] https://issues.apache.org/jira/browse/BEAM-12201
>>
>> On Wed, Apr 7, 2021 at 9:02 PM Wenbing Bai 
>> wrote:
>>
>>> Thank you, Brian. I tried `partition_cols`, but it is not working. I
>>> tried pure pandas, it does work, so I am not sure if anything wrong with
>>> Beam.
>>>
>>> Wenbing
>>>
>>> On Wed, Apr 7, 2021 at 2:56 PM Brian Hulette 
>>> wrote:
>>>
 Hm, to_parquet does have a `partition_cols` argument [1] which we pass
 through [2]. It would be interesting to see what  `partition_cols='key1'`
 does - I suspect it won't work perfectly though.

 Do you have any thoughts here Robert?

 [1]
 https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_parquet.html
 [2]
 https://github.com/apache/beam/blob/a8cd05932bed9b2480316fb8518409636cb2733b/sdks/python/apache_beam/dataframe/io.py#L525

 On Wed, Apr 7, 2021 at 2:22 PM Wenbing Bai 
 wrote:

> Hi Robert and Brian,
>
> I tried groupby in my case. Here is my pipeline code. I do see all the
> data in the final parquet file are sorted in each group. However, I'd like
> to write each partition (group) to an individual file, how can I achieve
> it? In addition, I am using the master of Apache Beam SDK, how can I test
> the pipeline with DataflowRunner considering there is no dataflow worker
> image available?
>
> data = [
> {
> "key1": 1000 + i % 10,
> "key2": randrange(1),
> "feature_1": "somestring{}".format(i)
> } for i in range(1)
> ]
>
> class TestRow(typing.NamedTuple):
> key1: int
> key2: int
> feature_1: str
>
> with beam.Pipeline() as p:
> pcoll = (
> p
> | beam.Create(data)
> | beam.Map(lambda x:x).with_output_types(TestRow)
> )
>
> df = to_dataframe(pcoll)
> sorted_df = df.groupby('key1').apply(lambda df: df.sort_values(by=
> 'key2')
> sorted_df.to_parquet('test_beam_dataframe{}.parquet'.format(str
> (uuid.uuid4())[:8]), engine='pyarrow', index=False)
>
> On Fri, Apr 2, 2021 at 10:00 AM Wenbing Bai 
> wrote:
>
>> Thank you, Robert and Brian.
>>
>> I'd like to try this out. I am trying to distribute my dataset to
>> nodes, sort each partition by some key and then store each partition to 
>> its
>> own file.
>>
>> Wenbing
>>
>> On Fri, Apr 2, 2021 at 9:23 AM Brian Hulette 
>> wrote:
>>
>>> Note groupby.apply [1] in particular should be able to do what you
>>> want, something like:
>>>
>>>   df.groupby('key1').apply(lambda df: df.sort_values('key2'))
>>>
>>> But as Robert noted we don't make any guarantees about preserving
>>> this ordering later in the pipeline. For this reason I actually just 
>>> sent a
>>> PR to disallow sort_values on the entire dataset [2].
>>>
>>> Brian
>>>
>>> [1] https://github.com/apache/beam/pull/13843
>>> [2] https://github.com/apache/beam/pull/14324